A few decades ago, buying a car meant one had to ensure some basic requirements were fulfilled – powerful engine capability, seating capacity, extent of service network, maybe features like a stereo and AC (depending on where you lived). Today, that checklist looks very different. We’re in a digital-first age and vehicles must be smart too. The concept of an intelligent, connected car was introduced by the automaking giant General Motors with the Cadillac Deville, Seville and Eldorado in 1996. By 2003, connected cars had a host of features like assisted driving, vehicle health checks and a network access device. Today, we have vehicles that work to build a digital persona of the driver and is perfectly personalized to echo his preferences to the tee.
A lot of this is made possible by the power of Artificial Intelligence. AI, and specifically deep learning, has become an important tool for the development of self-driving vehicles, particularly because of its ability to recognize and handle the nearly infinite number of scenarios encountered on the road. An intricate network of sensors, 2D & 3D LiDAR cameras and on-device computing is allowing vehicles to make decisions on the go, and intuitively guide one’s driving experience. This network of synchronized hardware and software can be delivered as a platform for vehicles. One of the most sought-after platforms is the Nvidia DRIVE. This computer platform designed by Nvidia is aimed at providing autonomous driving and car assistance, using deep learning. The platform was first introduced at the Consumer Electronics Show, Las Vegas in 2015. Since then, Nvidia has introduced annual upgrades with varied GPU architecture such as Drive CX & PX in 2016 (Maxwell-based), Drive PX2 for Tesla & Autochauffeur in 2016 & ‘17 (Pascal-based), Drive PX Xavier & Drive PX Pegasus in 2017 (Volta-based) and Drive AGX Orin in 2019 (Ampere-based).
What does Nvidia DRIVE offer?
First, the hardware. The embedded supercomputing platforms offered by DRIVE AGX process data from camera, radar, and lidar sensors to perceive the surrounding environment, localize the car to a map, and plan and execute a safe path forward. This AI platform supports autonomous driving, in-cabin functions and driver monitoring, as well as other safety features. Leading hardware offerings are the Nvidia DRIVE Orin, a System on Chip delivering 254 TOPS* and the central computer for intelligent vehicles. It is scalable from Level 2 to Level 5 autonomous vehicles. Nvidia DRIVE AGX Pegasus with two Xavier SoCs and two NVIDIA Turing GPUs to achieve 320 TOPS is built for Level 4 & Level 5 autonomous vehicles, including robotaxis. Nvidia DRIVE AGX Xavier, with 30 TOPS, is suited for Level 2 & Level 3 autonomous driving.
Other essential hardware include Nvidia DRIVE Hyperion. It is a reference and testing platform for autonomous vehicles with 12 exterior cameras, three interior cameras, nine radars, and two lidar sensors—and the Orin-based AI computing platform. It also features a full software stack for autonomous driving, driver monitoring, and visualization, and can be integrated into a test vehicle. And finally, there’s the Nvidia DRIVE Atlan, a 1,000 TOPS SoC, is a secure computing platform for AV development. Atlan was launched at Nvidia GTC 2021, touted to be a perpetually upgradable architecture through secure, over-the-air updates and is seen to be a long-term player for AV architecture.
Then comes the software. Software is what makes a machine intelligent. The open Nvidia DRIVE software stack allows developers to build state of the art AV applications including perception, localization and mapping, planning and control, driver monitoring, and natural language processing. Some of the DRIVE software are the DRIVE OS, DriveWorks, Drive AV and Drive IX, richly supported by a DNN framework and AR/VR visualization among others to support simulation and extensive data analysis.
This extensive hardware and software stack is complemented with the Nvidia DRIVE infrastructure encompasses the complete data center hardware, software, and workflows needed to develop and validate autonomous driving technology—from raw data collection through validation. It provides the building blocks required for neural network development, training and validation, replay, and testing in simulation.
*(Tera Operations Per Second)
Who uses Nvidia DRIVE?
Question is, who doesn’t? Nvidia believes that revolutionizing transport takes elaborate teamwork. The company works with more than 370 global automakers, Tier 1 suppliers, developers and researchers – all of whom are working in cohesion to integrating GPU technology and AI to transform deep learning, natural language processing, and gesture control technologies that will change how people drive, and how vehicles learn to drive themselves.
Nvidia is currently working with auto majors like Audi, Chery, Mercedes-Benz, Roborace, Tesla, Toyota, Uniti, Volkswagen and Volvo. Truck makers associated with Nvidia are DAF, Deutsche Post DHL Group, Einride, FAW, Iveco, Kenworth, Locomation, Navistar, NuPort Robotics, Peterbilt, Plus, Pony.ai, TuSimple and Volvo Autonomous Solutions. Given Nvidia’s diverse capabilities to cater to robotaxis, companies like Cruise, DiDi, Zoox, Optimus Ride and AutoX are also incorporating Nvidia’s DRIVE platform. Companies in the simulation, software, sensors and mapping sectors are also collaborating with Nvidia to support product development for the DRIVE platform.
Researchers at MIT recently developed a single deep neural network (DNN) for autonomous power vehicles where they used NVIDIA DRIVE AGX Pegasus to run the vehicle’s network, processing large amounts of LiDAR data in real-time.
During the launch of Nvidia DRIVE AGX Orin in December 2019, Sam Abuelsamid, principal research analyst at Navigant Research, said, “Nvidia’s long-term commitment to the transportation industry, along with its innovative end-to-end platform and tools, has resulted in a vast ecosystem — virtually every company working on AVs is utilizing NVIDIA in its compute stack.”
One of the most exciting aspects of Nvidia that draws the attention of leading car makers is their unquestionable superiority in software for automobiles. Danny Shapiro, Senior Director – Automotive, Nvidia noted: “We are not Tier Ones, and we will never replace them. Nonetheless, a lot of features and functions will be offered to vehicles in the form of software upgrades instead of gradual hardware replacements.”
Notably renowned for its GPUs and cloud computing, NVIDIA is unquestionably the market leader for AI computing in cars.
Source: indiaai.gov.in
